Small-signal measurements are affected by larger sigals on neighbouring channels of NI4351 DAQ

[John]

Hi,

I'm using LabWindows CVI V5.5.1 on Win 98 with an NI 4351 (24-bit) DAQ
and the TBX-68T accessory to measure temperature (with a K-type
thermocouple) and voltages, all on differential channels. I'm using
the NI435x CVI driver to set the appropriate measurement range for
each channel in a scan at the slow acquisition rate, with ground
referencing and cold junction compensation enabled. I need to measure
microvolts (dropped across a 10 ohm current-sense resistor for uA) on
some channels and much higher values (up to +/-15V) on others. However
I find that microvolt readings are badly affected when the higher
voltage readings on neighbouring channels increase. For example, a
200uV
reading with no large voltages on other channels changes to
800uV with 10V applied to any other channel. A temperature reading of
24°C changes to over 25°C with 10V applied and to under 23°C with
-10V. The level of deviation seems to be a function of the magnitude
of these larger voltages and is unaffected by averaging, as the whole
'floor' seems to shift. Tying 10K resistors from the positive
terminal of the temperature and small signal channels to analogue
ground improves things slightly, but the error is still too high.
Enabling auto-zero for every scan yielded no significant improvements.
I'd be grateful for any other suggestions.

Thanks in advance,
John.

[Reed_Blake]

John wrote in message <4ea5ba23.0409290205.1178aa0a@posting.google.com>...
>Hi,
>
>I'm using LabWindows CVI V5.5.1 on Win 98 with an NI 4351 (24-bit) DAQ
>and the TBX-68T accessory to measure temperature (with a K-type
>thermocouple) and voltages, all on differential channels. I'm using
>the NI435x CVI driver to set the appropriate measurement range for
>each channel in a scan at the slow acquisition rate, with ground
>referencing and cold junction compensation enabled. I need to measure
>microvolts (dropped across a 10 ohm current-sense resistor for uA) on
>some channels and much higher values (up to +/-15V) on others. However
>I find that microvolt readings are badly affected when the higher
>voltage readings on neighbouring channels increase.


Hi John,
I
have seen similar problems a few times. In one case it was bad hardware,
in the others
it had to do with the phenomenon of charge injection. NI has a knowledge
base article
0GJ8PJ03 that covers this issue. They also have an article titled
"Using a Unity Gain Buffer (Voltage Follower) with a DAQ Device"
in the Sensors and Transducers section of the development library.

The best solution I have found is to use an amplifier for each channel. I
usually specify
SCC carriers and modules for small systems.

I have also had some success with arranging the channels in decreasing gain
order
(low voltage channels first), then scanning a channel with shorted inputs
either
first or last. For example:

Ch Signal level
0 shorted input
1 low
2 low
3 low
4 medium
5 medium
6 high
7 high

Finally, scan at as slow a rate as possible.

Hopes this helps. This was a source of considerable frustration for me a few
years ago.
Regards,
Reed.
--
Reed Blake
Beta Technology
Indus
trial and Scientific Computing